Epithelial ovarian cancer (EOC) is the sixth most common cancer in women worldwide and, despite advances in detection and therapies, it still represents the most lethal gynecologic malignancy in the industrialized countries. Unfortunately, still relatively little is known about the molecular events that lead to the development of this highly aggressive disease. The relatively recent discovery of microRNAs (miRNA), a class of small noncoding RNAs targeting multiple mRNAs and triggering translation repression and/or RNA degradation, has revealed the existence of a new level of gene expression regulation. Multiple studies involving various types of human cancers proved that miRNAs have a causal role in tumorigenesis. Here we show that, in comparison to normal ovary, miRNAs are aberrantly expressed in human ovarian cancer. The overall miRNA expression could clearly separate normal versus cancer tissues. The most significantly overexpressed miRNAs were miR-200a, miR-141, miR-200c, and miR-200b, whereas miR-199a, miR-140, miR-145, and miR-125b1 were among the most down-modulated miRNAs. We could also identify miRNAs whose expression was correlated with specific ovarian cancer biopathologic features, such as histotype, lymphovascular and organ invasion, and involvement of ovarian surface. Moreover, the levels of miR-21, miR-203, and miR-205, up-modulated in ovarian carcinomas compared with normal tissues, were significantly increased after 5-aza-2′-deoxycytidine demethylating treatment of OVCAR3 cells, suggesting that the DNA hypomethylation could be the mechanism responsible for their overexpression. Our results indicate that miRNAs might play a role in the pathogenesis of human EOC and identify altered miRNA gene methylation as a possible epigenetic mechanism involved in their aberrant expression. [Cancer Res 2007;67(18):8699–707]

Epithelial ovarian cancer (EOC) is the most common gynecologic malignancy and the sixth most common cancer in women worldwide, with highly aggressive natural history causing almost 125,000 deaths yearly (1). Despite advances in detection and cytotoxic therapies, only 30% of patients with advanced-stage ovarian cancer survive 5 years after initial diagnosis (2). The high mortality of this disease is mainly due to late-stage diagnosis for >70% of ovarian cancers. In fact, when ovarian cancer is diagnosed in its early stage, that is still organ confined, the 5-year survival rate exceeds 90%. Unfortunately, only 19% of all ovarian cancers are diagnosed at this early stage. Indeed, this rather poor prognosis is due to (a) the insidious asymptomatic nature of this disease in its early onset, (b) the lack of robust and minimally invasive methods for early detection, and (c) tumor resistance to chemotherapy. The vast majority of human ovarian carcinomas are represented by ovarian epithelial cancers (OEC), deriving from the ovarian surface epithelium (3).

Ovarian adenocarcinomas occur as four major histologic subtypes, serous, mucinous, endometrioid, and clear cell, with serous being the most common. Current data indicate that each of these histologic types is associated with distinct morphologic and molecular genetic alterations (4), but further investigations of the molecular mechanisms promoting ovarian cancer are necessary to determine how each of the subtypes emerges.

Over the last 5 years, expression profiling technologies greatly improved, thus expanding the knowledge on cancer etiology and biomarkers with clinical applications (5, 6). However, although these technologies have provided most of the new biomarkers with potential use for diagnosis, drug development, and tailored therapy, they have thus far shed little insight into the detailed mechanisms at the origin of this neoplasia, thus suggesting that ovarian tumorigenesis may occur through novel or poorly characterized pathways.

A new class of small noncoding RNAs, named microRNAs (miRNA), was discovered recently and shown to regulate gene expression at post-transcriptional level, for the most part by binding through partial sequence homology to the 3′ untranslated region of target mRNAs and causing block of translation and/or mRNA degradation (7). miRNAs are 19 to 25 nt long molecules cleaved from 70 to 100 nt hairpin pre-miRNA precursors. The precursor is cleaved by cytoplasmic RNase III Dicer into ∼22 nt miRNA duplex: one strand (miRNA*) of the short-lived duplex is degraded, whereas the other strand, which serves as mature miRNA, is incorporated into the RNA-induced silencing complex and drives the selection of target mRNAs containing antisense sequences.

Several studies have shown that miRNAs play important roles in essential processes, such as differentiation, cell growth, and cell death (8, 9).

Moreover, it has been shown that miRNAs are aberrantly expressed or mutated in cancers, suggesting that they may play a role as a novel class of oncogenes or tumor suppressor genes, depending on the targets they regulate: let-7, down-regulated in lung cancer, suppresses RAS (10) and HMGA2 (11, 12); mir-15 and mir-16, deleted or down-regulated in leukemia, suppress BCL2 (13); and mir-17-5p and mir-20a, control the balance of cell death and proliferation driven by the proto-oncogene c-Myc (14). Clear evidences indicate that miRNA polycistron mir-17-92 acts as an oncogene in lymphoma and lung cancer (15); mir-372 and mir-373 are novel oncogenes in testicular germ cell tumors by numbing p53 pathway (16); miR-155, overexpressed in B-cell lymphomas and solid tumors, leads to the development of B-cell malignancies in an in vivo model of transgenic mice (17).

The use of miRNA microarray technologies has been used as a powerful tool to recognize miRNAs differentially expressed between normal and tumor samples (1820) and also to identify miRNA expression signatures associated with well-defined clinicopathologic features and disease outcome (21, 22). Several studies have also investigated the molecular mechanisms leading to an aberrant miRNA expression, identifying the presence of genomic abnormalities in miRNA genes (21, 23, 24). More recently, few evidences have shown that miRNA genes may be regulated also by epigenetic mechanisms, as changes in genomic DNA methylation pattern: miR-127 (25) and miR-124a (26) are transcriptionally inactivated by CpG island hypermethylation, whereas in lung cancer, the overexpression of let-7a-3 seems to be due to DNA hypomethylation (27).

Here, we present the results of a genome-wide miRNA expression profiling in a large set of normal and tumor ovarian tissues, showing the existence of an ovarian cancer-specific miRNA signature and identifying the altered methylation of miRNA genes as a possible epigenetic mechanism responsible for their aberrant expression.

Ovarian cancer samples and cell lines. A total of 84 snap-frozen normal and malignant ovarian tissues were collected at the GOG Tissues Bank, Columbus Children's Hospital (Columbus, OH). The tissue collection used for microarray analysis included 15 normal ovarian tissue sections and 69 malignant tissues, all ovarian epithelial carcinomas, including 31 serous (29 of them showed a papillary pattern), 8 endometrioid, 4 clear cell, 9 poorly differentiated, and 1 mucinous carcinomas. The ovarian cancer cell line IGROV1 was originally derived by Dr. Bernard (Institute Gustave Roussy, Villejuf, France), from a moderately differentiated ovarian carcinoma of an untreated patient; OAW-42 was from Dr. Ulrich U. (Department of Obstetrics and Gynaecology, University of Ulm, Ulm, Germany), whereas OVCAR3, OVCAR8, and SKOV3 were purchased from the American Type Culture Collection. All the cell lines were maintained in RPMI 1640 (Life Technologies), supplemented with 10% (v/v) fetal bovine serum, 3 mmol/L l-glutamine, and 100 units/mL penicillin/streptomycin.

miRNA microarray hybridization and quantification. Total RNA isolation was done with Trizol (Invitrogen) according to the manufacturer's instructions. RNA labeling and hybridization on miRNA microarray chips were done as described previously (28) using 5 μg of total RNA from each sample. Hybridization was carried out on our miRNA microarray (Ohio State Comprehensive Cancer Center, version 2.0), which contains probes for 460 mature miRNAs spotted in quadruplicate (235 Homo sapiens, 222 Mus musculus, and three Arabidopsis thaliana) with annotated active sites. Often, more than one probe set exists for a given mature miRNA. Additionally, there are quadruplicate probes corresponding to most precursor miRNAs. Hybridization signals were detected with streptavidin-Alexa Fluor 647 conjugate and scanned images (Axon 4000B) were quantified using the GenePix 6.0 software (Axon Instruments).

Computational analysis of miRNA microarray data. Microarray images were analyzed by using GenePix Pro. Average values of the replicate spots of each miRNA were background subtracted, normalized, and subjected to further analysis. We did a global median normalization of ovary microarray data by using BRB ArrayTools developed by Richard Simon and Amy Peng Lam (29). Absent calls were thresholded to 4.5 before subsequent statistical analysis. This level is the average minimum intensity level detected in the experiments. miRNA nomenclature was according to the Genome Browser3

and the miRNA database at Sanger Center4; in case of discrepancies, the miRNA database was followed. Differentially expressed miRNAs were identified by using the t test procedure within significance analysis of microarrays (SAM), a method developed at Stanford University Labs based on recent article of Tusher, Tibshirani, and Chu (30). To identify miRNA signatures, we also applied PAM, which does sample classification from gene expression data, via the “nearest shrunken centroid method” of Tibshirani, Hastie, Narasimhan, and Chu (31).

Northern blotting. Northern blot analysis was done as described previously. RNA samples (10 μg each) were run on 15% polyacrylamide and 7 mol/L urea Criterion precasted gels (Bio-Rad) and transferred onto Hybond-N+ membranes (Amersham). The hybridization was done at 37°C in ULTRAhyb-Oligo hybrization buffer (Ambion) for 16 h. Membranes were washed at 37°C, twice with 2× saline-sodium phosphate-EDTA and 0.5% SDS. The oligonucleotides used as probes were antisense to the sequence of the mature miRNAs (miRNA Registry5

): miR-200a, 5′-ACATCGTTACCAGACAGTGTTA-3′; miR-141, 5′-CCATCTTTACCAGACAGTGTTA-3′; miR-199a, 5′-GAACAGGTAGTCTGAACACTGGG-3′; miR-125b1, 5′-TCACAAGTTAGGGTCTCAGGGA-3′; miR-145, 5′-AAGGGATTCCTGGGAAAACTGGAC-3′; miR-222, 5′-GAGACCCAGTAGCCAGATGTAGCT-3′; and miR-21, 5′-TCAACATCAGTCTGATAAGCTA-3′.

5S RNA or EtBr gel staining were used to normalize. Two hundred nanograms of each probe were end labeled with 100 μCi [γ-32P]ATP using the polynucleotide kinase (Roche). Blots were stripped in boiling 0.1% SDS for 10 min before rehybridization.

Quantitative real-time PCR. The single tube Taqman miRNA assays were used to detect and quantify mature miRNAs on Applied Biosystems real-time PCR instruments in accordance with manufacturer's instructions (Applied Biosystems). Normalization was done with 18S rRNA. All reverse transcription reactions, including no-template controls and reverse transcriptase minus controls, were run in a GeneAmp PCR 9700 Thermocycler (Applied Biosystems). Gene expression levels were quantified using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Comparative real-time PCR was done in triplicate, including no-template controls. Relative expression was calculated using the comparative Ct method.

Demethylating experiment. OVCAR3 cells were seeded at low density 48 h before treatment with 10 μmol/L 5-aza-2′-deoxycytidine (5-AZA; Sigma). After 24 h of treatment, cells were collected and total RNA was isolated using Trizol reagent. Three replicates for both untreated cells and 5-AZA–treated cells were used to evaluate the miRNA expression by microarray profiling. Differentially expressed miRNAs were identified by using univariate two-class t test with random variance model.

A miRNA expression signature discriminates ovarian cancer tissues from normal ovary. We used a custom microarray platform already validated by numerous studies (19) to evaluate miRNA expression profiles on a heterogeneous set of ovarian tissues from different patients. This set included 15 normal ovarian samples, 69 ovarian malignant tumors, and 5 ovarian cancer cell lines, for a total of 89 biologically independent samples. Each tumor sample was derived from a single specimen. Description of sample characteristics is reported in Supplementary Table S1.

The unsupervised hierarchical clustering, based on all the human miRNAs spotted on the chip, generated a tree with a clear distinction of samples in two main groups, represented by normal tissues and malignant tissues (Fig. 1).

To identify miRNAs differentiating normal versus cancer tissue, we used SAM and PAM tools, and the results obtained from the two types of class prediction analysis were largely overlapping. The SAM comparison between normal and cancer tissues identified 39 miRNAs (with q values <1% and fold changes >3) differentially expressed, 10 up-modulated in tumors and the remaining down-modulated (the list is reported in Supplementary Table S2). The PAM analysis in Supplementary Fig. S1 displays the graphical representation of the probabilities (0.0–1.0) of each sample for being a cancer or a normal tissue according to the miRNA signature reported in Table 1, which describes a smaller set of 29 miRNAs, 4 up-modulated (miR-200a, miR-200b, miR-200c, and miR-141) and 25 down-modulated (being miR-199a, miR-140, miR-145, and miR-125b1 among the most significant) differentiating normal versus tumor with a classification rate of 89%.

To confirm the results obtained by microarray analysis, we carried out Northern blots (Fig. 2A) or real-time PCR (Fig. 2B) on some of the differentially expressed miRNAs. We analyzed the expression of miR-200a and miR-141, the most significantly up-modulated in ovarian carcinoma, and the miRNAs most significantly down-modulated, miR-199a, miR-140, miR-145, and miR-125b1. All the experiments confirmed the results obtained by microarray analysis.

Biopathologic features and miRNA expression. Considering that ovarian epithelial carcinomas occur as different histologic subtypes characterized by distinct morphologic and molecular genetic alterations, we decided to compare the miRNA profile of each of them with the normal tissue to evaluate if miRNA expression profiles are different in distinct histotypes of ovarian carcinomas. Complete lists resulting from SAM analyses are reported in Supplementary Table S3, whereas a summary is shown in the Venn diagram in Fig. 3: two of four miRNAs most significantly up-modulated (Fig. 3A) in tumors versus normal tissue, miR-200a and miR-200c, are up-modulated in all the three histotypes considered (serous, endometrioid, and clear cell), whereas miR-200b and miR-141 up-modulation is shared by endometrioid and serous histotypes. Moreover, the endometrioid histotype shows the up-modulation of three additional miRNAs, miR-21, miR-203, and miR-205. Nineteen miRNAs, including miR-125b1, miR199a, and miR-140, are down-modulated (Fig. 3B) in all the three histotypes examined in comparison with normal tissue, whereas four are shared in each paired analysis of the different signatures: miR-145, for example, is down-modulated in both serous and clear cell carcinomas; miR-222 is down-modulated in both endometrioid and clear cell carcinomas.

Considering the tumors classified as “mixed” and “poorly differentiated,” we found that the first group revealed a signature with characteristics of different histotypes, sharing for example the overexpression of miR-200c and miR-181 with the endometrioid carcinomas and the down-modulation of miR-214 with the serous, whereas the “poorly differentiated” tumors have a quite different pattern of miRNA expression (Supplementary Table S3).

We then compared miRNA expression of different groups of tumors paired as reported in Supplementary Table S4, and in particular, we compared the two most numerous histotypes, serous and endometrioid. When considering the miRNAs differentially expressed in endometrioid carcinomas compared with serous, we found miR-212 up-modulated and miR-302b* and miR-222 (P < 0.05, t test analysis of microarray data in Fig. 4A) among the miRNAs most significantly down-modulated. In Fig. 4B, a Northern blot on a small set of samples verifies miR-222 overexpression in serous tumors compared with endometrioid.

We then focused our attention on other clinicopathologic features associated with tumor specimens: whereas no miRNAs were found significantly differentially expressed in relation to the age of patients, other tumor characteristics seemed to affect miRNA expression, such as lymphovascular invasion and ovarian surface, tubal, uterus, and pelvic peritoneum involvement (Supplementary Table S5).

To investigate if there were miRNAs associated with different grade or stage of the disease, we did comparative analyses considering all the tumors or only the serous histotype, which was the most numerous, but we did not obtain any significant miRNA differentially expressed.

Confirmed and potential targets for miRNA members of various signatures. Using the DianaTarbase,6

we looked for confirmed targets of some of the most significant miRNAs resulting from our analyses, finding some interesting data: ERBB2 and ERBB3 receptors, for example, are targeted by miR-125 (32); miR-101, down-modulated in ovarian carcinoma, has been shown targeting the oncogene MYCN (33). We then analyzed their potential targets using the miRGen database7 and evaluated for some of these molecules the expression levels in ovarian carcinoma: all the four most significantly up-modulated miRNAs, miR-200a, miR-200b, miR-200c, and miR-141, for example, have as common putative target the oncosuppressor BAP1, BRCA1-associated protein, down-modulated in ovarian cancer. The information obtained is summarized in Supplementary Table S6.

Epigenetic regulation of miRNA expression. To evaluate if an aberrant DNA methylation pattern could also contribute to the altered miRNA expression characterizing the human ovarian carcinoma, we analyzed the miRNA profiling of the ovarian cell line OVCAR3, before and after treatment with the demethylating agent 5-AZA. The analysis of microarray data showed 11 human miRNAs differentially expressed, 9 up-modulated and 2 down-modulated (P < 0.001, significance threshold of each univariate test), being miR-21, miR-203, miR-146b, miR-205, miR-30-5p, and miR-30c the most significantly induced on treatment (the miRNAs differentially expressed are listed in Fig. 5A, whereas the resulting hierarchical cluster tree is reported in Fig. 5B). Real-time PCRs to verify the up-modulation of the five most significantly induced miRNAs are described in Fig. 5 as graphical representation of miRNA expression levels (Fig. 5C), and miR-21 was also validated by Northern blot (Fig. 5D).

Interestingly, miR-21, miR-203, and miR-205 are overexpressed in ovarian carcinomas compared with normal tissues (see SAM analysis in Supplementary Table S2 and Venn Diagram in Fig. 3): the reactivation of these miRNA genes after demethylating treatment suggests that the hypomethylation could be the mechanism responsible for their overexpression in vivo. We confirmed the overexpression of miR-21, the most significant miRNA induced on treatment, doing a Northern blotting (Supplementary Fig. S2A) on a panel of human ovarian carcinomas and two normal tissues. Moreover, using the CpG Island Searcher Program (34), we verified that miR-21 and miR-203 are associated with CpG islands, being the miR-203 embedded in a CpG island 875 bp long and the miR-21 characterized by a CpG island ≈2 kb upstream the mature sequence (Supplementary Fig. S2B), whereas miR-205 does not show any CpG island in a region spanning 2 kb upstream its mature form.

In this study, we show that miRNAs are aberrantly expressed in human ovarian cancer. The overall miRNA expression could clearly separate normal versus cancer tissues, identifying a number of miRNAs altered in human ovarian cancer and probably involved in the development of this neoplasia.

The expression of all the four miRNAs we found most significantly up-modulated, miR-200a and miR-141, belonging to the same family; miR-200b (localized in the same region of miR-200a, at chromosome 1p36.33); and miR-200c (localized in the same region of miR-141, at chromosome 12p13.31), is concordant with the results obtained at genomic level by Zhang et al. (24), suggesting that the mechanism driving their up-modulation could be the amplification of the miRNA genes. Interestingly, all these miRNAs have a common putative target: the oncosuppressor BAP1, BRCA1-associated protein (24). The altered expression of GATA factors, found and proposed as the underlying mechanism for dedifferentiation in ovarian carcinogenesis (35), may also be driven by miRNA deregulation. In particular, GATA6, lost or excluded from the nucleus in 85% of ovarian tumors, may be regulated by miR-200a, and GATA4, absent in the majority of ovarian cancer cell lines, may be targeted by miR-200b (Supplementary Table S5).

Among the down-modulated genes, notably, we found miR-125b1, altered also in breast cancer, as well as miR-145 (18); mir-199a, recently shown down-modulated in other tumors, as hepatocellular carcinoma (36); and miR-140, deleted in ovarian carcinoma (24). Interestingly, miR-140 is indeed located at chromosome 6q22, a fragile region often deleted in ovarian tumor, and it is predicted to target important molecules as c-SRK, MMP13, and FGF2.

Even if the normal control available in this study is represented by whole normal ovary, our data could identify a number of miRNAs altered in human ovarian carcinoma and probably involved in the biology of this malignancy.

In fact, the miRNA signatures obtained comparing different histotypes of ovarian carcinomas (serous, endometrioid, clear cell, and mixed) with the normal tissue are overlapping in most cases, but they also reveal a number of miRNAs that seem to be “histotype specific”: the endometrioid tumors, for example, share with the others the four most significantly up-modulated miRNAs (miR-200a, miR-200b, miR-200c, and miR-141) but also present overexpression of miR-21, known to be misregulated in numerous solid tumors (18, 37, 38) and to exert an antiapoptotic role in different cellular systems (39, 40), miR-205 and miR-182. Endometrioid tumors also present down-modulation of several miRNAs in comparison with the other classes of tumors, for example, miR-222, already shown targeting c-Kit (41), being involved in cancer (4244) and down-modulated under folate-deficient conditions (45).

These differences enforce the fact that different histotypes represent biologically and pathogenetically distinct entities of EOCs, although they are currently treated with identical therapeutic strategies. Microarray analysis has confirmed recently that different histotypes (serous, mucinous, endometrioid, and clear cell) show the alteration of different pathways, probably reflecting the gene expression pattern of the organ of origin (respectively fallopian tubes, colonic mucosa, and endometrium; ref. 46).

Notably, many of the miRNAs differentially expressed are predicted to target molecules involved in pathways differentially activated depending on the histotype. miR-212, for example, down-modulated in serous carcinoma, has as putative target WT1, overexpressed in this subgroup of ovarian carcinomas (47). Another putative target of miR-212 is BRCA1: mutated in hereditary ovarian cancer, this molecule has been found recently involved also in the pathoetiology of sporadic OEC, where a loss of gene function due to epigenetic alterations has been observed more commonly (48). The decreased BRCA1 expression could be determined by overexpression of one or more miRNAs.

miR-299-5p and miR-135b, up-modulated in serous histotype compared with endometrioid, are supposed to target, respectively, Delta-like 1 (DLK1) and msh homeobox 2 (MSX2), overexpressed in endometrioid carcinomas (47). Compared with the other tumors, clear cell carcinomas show expression levels of miR-30-5p and of miR-20a opposite (46) to two putative targets, retinol binding protein 4 (RBP4) and solute carrier 40-iron-regulated transporter, member 1 (SLC40A1), respectively. Compared with the normal tissue, clear cell carcinoma also shows lower expression of miR-18a, miR-19a, and miR-19b, suggesting a possible down-modulation of the cluster 17-92 (already validated as deleted by Zhang et al.). This cluster, involved in the intricate regulation mediated by E2F1 and c-Myc, seems to have a duplex nature of putative oncogene, as suggested recently in B-cell lymphoma (15), or tumor suppressor: in hepatocellular carcinoma, for example, loss of heterozygosity at the locus coding the miR-17-92 cluster (13q31) has been reported (49). In ovarian carcinoma, at least in clear cell histotype, it could also exert a role of oncosuppressor. Our data suggest indeed that miRNAs may have a regulatory role in the process of differentiation leading to the development of a specific subtype of EOC. Interestingly, poorly differentiated carcinomas have a quite different pattern of miRNA expression, showing up-modulation of several miRNAs in comparison with normal ovary. More intriguingly, one of them, miR-373, has been described recently as putative oncogene in testicular germ cell tumors (16).

The absence of miRNAs significantly differentially expressed in relation to tumor stage or grade might be explained by the fact that our set of samples is mostly represented by advanced stage tumors, as expected considering the late diagnosis of this kind of neoplasia; however, the difference in size among the different groups of samples could have represented a limit for the statistical analysis. Alternatively, miRNAs might be important for the development of human ovarian carcinoma but not for the progression of the disease.

Resulting from our analyses, a number of miRNAs overexpressed but not reported as amplified in Zhang study, as well as down-modulated but not deleted, the involvement of an epigenetic regulatory mechanism could actually exert a role on miRNA expression in human EOC. Indeed, among the most significant miRNAs induced after demethylating treatment of an ovarian cell line, we found miR-21, miR-203, and miR-205 up-modulated in ovarian cancer. Moreover, miR-203 and miR-21 are associated with a CpG island (miR-203 is embedded in a CpG island, whereas miR-21 has a CpG island ≈2 kb upstream its mature sequence), supporting the idea that the demethylation leads to the reactivation of these miRNA genes. Notably, miR-21 has already been described up-modulated in several human tumors and having an antiapoptotic role in different cellular models. These data suggest that the DNA hypomethylation could be an epigenetic mechanism responsible for the in vivo overexpression of potentially oncogenic miRNAs.

To the best of our knowledge, this is the first report describing a complete miRNA expression profiling in human EOCs, focused on the identification of miRNAs differentially expressed in carcinomas versus normal ovary and in different subgroups of tumors. Our data suggest the important role that miRNAs can exert on the pathogenesis and on the development of different histotypes of ovarian carcinoma and identify altered DNA methylation as a possible epigenetic mechanism responsible for the aberrant expression of miRNAs not affected by genomic changes.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

Grant support: Associazione Italiana per la Ricerca sul Cancro. Fondazione Italiana per la Ricerca sul Cancro fellowship (M.V. Iorio). National Cancer Institute Program Project grants (C.M. Croce). Kimmel Foundation Scholar award and CLL Global Research Foundation (G.A. Calin).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Elda Tagliabue for the scientific discussion and her useful advice.

1
Cannistra SA. Cancer of the ovary.
N Engl J Med
2004
;
351
:
2519
–29.
2
Greenlee RT, Hill-Harmon MB, Murray T, Thun M. Cancer statistics, 2001.
CA Cancer J Clin
2001
;
51
:
15
–36.
3
Feeley KM, Wells M. Precursor lesions of ovarian epithelial malignancy.
Histopathology
2001
;
38
:
87
–95.
4
Bell DA. Origins and molecular pathology of ovarian cancer.
Mod Pathol
2005
;
18
Suppl 2:
S19
–32.
5
Schwartz DR, Kardia SL, Shedden KA, et al. Gene expression in ovarian cancer reflects both morphology and biological behavior, distinguishing clear cell from other poor-prognosis ovarian carcinomas.
Cancer Res
2002
;
62
:
4722
–9.
6
De Cecco L, Marchionni L, Gariboldi M, et al. Gene expression profiling of advanced ovarian cancer: characterization of a molecular signature involving fibroblast growth factor 2.
Oncogene
2004
;
23
:
8171
–83.
7
He L, Hannon GJ. MicroRNAs: small RNAs with a big role in gene regulation.
Nat Rev Genet
2004
;
5
:
522
–31.
8
Miska EA. How microRNAs control cell division, differentiation, and death.
Curr Opin Genet Dev
2005
;
5
:
563
–8.
9
Zamore PD, Haley B. Ribo-gnome: the big world of small RNAs.
Science
2005
;
309
:
1519
–24.
10
Johnson SM, Grosshans H, Shingara J, et al. RAS is regulated by the let-7 microRNA family.
Cell
2005
;
120
:
635
–47.
11
Mayr C, Hemann MT, Bartel D. Disrupting the pairing between let-7 and HMGA2 enhances oncogenic transformation.
Science
2007
;
315
:
1576
–9.
12
Lee YS, Dutta A. The tumor suppressor microRNA let-7 represses the HMGA2 oncogene.
Genes Dev
2007
;
21
:
1025
–30.
13
Cimmino A, Calin GA, Fabbri M, et al. miR-15 and miR-16 induce apoptosis by targeting BCL2.
Proc Natl Acad Sci U S A
2005
;
102
:
13944
–9.
14
O'Donnell KA, Wentzel EA, Zeller KI, Dang CV, Mendell JT. c-Myc-regulated microRNAs modulate E2F1 expression.
Nature
2005
;
435
:
839
–43.
15
He L, Thomson JM, Hemann MT, et al. A microRNA polycistron as a potential human oncogene.
Nature
2005
;
435
:
828
–33.
16
Voorhoeve PM, le Sage C, Schrier M, et al. A genetic screen implicates miRNA-372 and miRNA-373 as oncogenes in testicular germ cell tumors.
Cell
2006
;
124
:
1169
–81.
17
Costinean S, Zanesi N, Pekarsky Y, et al. Pre-B cell proliferation and lymphoblastic leukemia/high-grade lymphoma in E(mu)-miR155 transgenic mice.
Proc Natl Acad Sci U S A
2006
;
103
:
7024
–9.
18
Iorio MV, Ferracin M, Liu CG, et al. MicroRNA gene expression deregulation in human breast cancer.
Cancer Res
2005
;
65
:
7065
–70.
19
Calin GA, Croce CM. MicroRNA signatures in human cancers.
Nat Rev Cancer
2006
;
6
:
857
–66.
20
Esquela-Kerscher A, Slack FJ. Oncomirs—microRNAs with a role in cancer.
Nat Rev Cancer
2006
;
6
:
259
–69.
21
Calin GA, Ferracin M, Cimmino A, et al. MicroRNA signature associated with prognosis and progression in chronic lymphocytic leukemia.
N Engl J Med
2005
;
353
:
1793
–801.
22
Yanaihara N, Caplen N, Bowman E, et al. Unique microRNA molecular profiles in lung cancer diagnosis and prognosis.
Cancer Cell
2006
;
9
:
189
–98.
23
Calin GA, Croce CM. MicroRNAs and chromosomal abnormalities in cancer cells.
Oncogene
2006
;
25
:
6202
–10.
24
Zhang L, Huang J, Yang N, et al. MicroRNAs exhibit high frequency genomic alterations in human cancer.
Proc Natl Acad Sci U S A
2006
;
103
:
9136
–41.
25
Saito Y, Liang G, Egger G, et al. Specific activation of microRNA-127 with downregulation of the proto-oncogene BCL6 by chromatin-modifying drugs in human cancer cells.
Cancer Cell
2006
;
9
:
435
–43.
26
Lujambio A, Ropero S, Ballestar E, et al. Genetic unmasking of an epigenetically silenced microRNA in human cancer cells.
Cancer Res
2007
;
67
:
1424
–9.
27
Brueckner B, Stresemann C, Kuner R, et al. The human let-7a-3 locus contains an epigenetically regulated microRNA gene with oncogenic function.
Cancer Res
2007
;
67
:
1419
–23.
28
Liu CG, Calin GA, Meloon B, et al. An oligonucleotide microchip for genomewide microRNA profiling in human and mouse tissues.
Proc Natl Acad Sci U S A
2004
;
101
:
9740
–4.
29
Wright GW, Simon RM. A random variance model for detection of differential gene expression in small microarray experiments.
Bioinformatics
2003
;
19
:
2448
–55.
30
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response.
Proc Natl Acad Sci U S A
2001
;
98
:
5116
–21.
31
Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression.
Proc Natl Acad Sci U S A
2002
;
99
:
6567
–72.
32
Scott GK, Goga A, Bhaumik D, et al. Coordinate suppression of ERBB2 and ERBB3 by enforced expression of micro-RNA miR-125a or miR-125b.
J Biol Chem
2007
;
282
:
1479
–86.
33
Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. Prediction of mammalian microRNA targets.
Cell
2003
;
115
:
787
–98.
34
Takai D, Jones PA. The CpG island searcher: a new WWW resource.
In Silico Biol
2003
;
3
:
325
–40.
35
Capo-chichi CD, Roland IH, Vanderveer L, et al. Anomalous expression of epithelial differentiation—determining GATA factors in ovarian tumorigenesis.
Cancer Res
2003
;
63
:
4967
–77.
36
Murakami Y, Yasuda T, Saigo K, et al. Comprehensive analysis of microRNA expression patterns in hepatocellular carcinoma and non-tumorous tissues.
Oncogene
2006
;
25
:
2537
–45.
37
Roldo C, Missaglia E, Hagan JP, et al. MicroRNA expression abnormalities in pancreatic endocrine and acinar tumors are associated with distinctive pathologic features and clinical behavior.
J Clin Oncol
2006
;
24
:
4677
–84.
38
Volinia S, Calin GA, Liu CG, et al. A microRNA expression signature of human solid tumors defines cancer gene targets.
Proc Natl Acad Sci U S A
2006
;
103
:
2257
–61.
39
Chan JA, Krichevsky AM, Kosik KS. MicroRNA-21 is an antiapoptotic factor in human glioblastoma cells.
Cancer Res
2005
;
65
:
6029
–33.
40
Zhu S, Si ML, Wu H, Mo YY. MicroRNA-21 targets the tumor suppressor gene tropomyosin 1 (TPM1).
J Biol Chem
2007
;
282
:
14328
–36.
41
Felli N, Fontana L, Pelosi E, et al. MicroRNAs 221 and 222 inhibit normal erythropoiesis and erythroleukemic cell growth via kit receptor downmodulation.
Proc Natl Acad Sci U S A
2005
;
102
:
18081
–6.
42
He H, Jazdzewski K, Li W, et al. The role of microRNA genes in papillary thyroid carcinoma.
Proc Natl Acad Sci U S A
2005
;
102
:
19075
–80.
43
Pallante P, Visone R, Ferracin M, et al. MicroRNA deregulation in human thyroid papillary carcinomas.
Endocr Relat Cancer
2006
;
13
:
497
–508.
44
Lee EJ, Gusev Y, Jiang J, et al. Expression profiling identifies microRNA signature in pancreatic cancer.
Int J Cancer
2007
;
120
:
1046
–54.
45
Marsit CJ, Eddy K, Kelsey KT. MicroRNA responses to cellular stress.
Cancer Res
2006
;
66
:
10843
–8.
46
Marquez RT, Baggerly KA, Patterson AP, et al. Patterns of gene expression in different histotypes of epithelial ovarian cancer correlate with those in normal fallopian tube, endometrium, and colon.
Clin Cancer Res
2005
;
11
:
6116
–26.
47
Shedden KA, Kshirsagar MP, Schwartz DR, et al. Histologic type, organ of origin, and Wnt pathway status: effect on gene expression in ovarian and uterine carcinomas.
Clin Cancer Res
2005
;
11
:
2123
–31.
48
Thrall M, Gallion HH, Kryshio R, et al. BRCA1 expression in a large series of sporadic ovarian carcinomas: a Gynecologic Oncology Group study.
Int J Gynecol Cancer
2006
;
16
Suppl 1:
166
–71.
49
Lin YW, Sheu JC, Liu LY, et al. Loss of heterozygosity at chromosome 13q in hepatocellular carcinoma: identification of three independent regions.
Eur J Cancer
1999
;
35
:
1730
–4.